Data-driven model-based instability prediction in an annular combustor relying on a flame response mapping of the operating domain

At a stage where new architectures and alternative fuels are being proposed to tackle the environmental challenges, it is important to be able to deal with combustion dynamics issues that may arise in these new developments. Reduced order models are generally considered for that purpose but their capacity to predict combustion instabilities is still not fully demonstrated. One advantage of these models is that they mainly rely on flame transfer or describing functions (FTFs or FDFs) representing the flames’ response to incoming disturbances. Recent measurements indicate that FDFs exhibit gain and phase variations with fuels, fuel blends, injector characteristics, but also with operating conditions. However, FTFs and FDFs are generally documented only for a few operating points and do not cover the entire domain of operation, limiting the scope of the analysis. The logical step taken in the present investigation is to collect FDFs for a large number of flow conditions of the laboratory-scale annular combustor MICCA-Spray. This is achieved using a single-injector system, SICCA-Spray, representing one sector of MICCA-Spray and that allows external flame modulation. The collected FDF data correspond to injectors of two types, characterized by different combustion dynamics in MICCA-Spray. This FDF database, in combination with an analytical framework derived from acoustic energy balance equations, serves to determine growth rates and define a theoretical instability domain. A comparison with the stability maps obtained in the annular combustor indicates that the general layout of these maps can be retrieved for the two injector types, validating the relevance of this data-driven model-based analysis of thermo-acoustic instabilities.

Novelty and significance statement

The novelty of this work lies in the reported flame describing function (FDF) database, measured in the single-injector setup SICCA-Spray, for a wide range of operating conditions corresponding to the operation domain of the MICCA-Spray annular combustor, and for two types of injectors leading to different flame dynamics (stable and unstable). An analytical framework is then used to determine growth rates of oscillation based on the FDF data, enabling to perform a stability analysis and interpret the observations in MICCA-Spray: the differences in flame dynamics observed between the two injectors are successfully retrieved, and for the unstable injector, stable and unstable regions of the operating domain can also be distinguished.

This work is significant because it provides an analytical framework of interest from a theoretical standpoint and for practical applications that is validated against a broad experimental dataset.